The point is that we need to be careful not to take this data as implying that games cause better performance, only that it's correlated with it; guelo pointed out the study does control for BMI/IQ/etc, but I think that this is still insufficient to claim any kind of causation there*. We know nothing at all of the counterfactual world where those same kids didn't play games.

* Even if IQ tests were a message from the heavens saying THESE KIDS WERE DESTINED FOR THE EXACT SAME LEVEL OF COGNITIVE PERFORMANCE AT BIRTH we would still know nothing about causality, because there may have been something else in their lives that caused both the game-playing and the better performance.